Cross-Validation of a Classification System for Persons With Traumatic Brain Injury in the Posthospital Period. Issue 5 (September 2017)

• Record Type: Journal Article
• Authors: Sherer, Mark; Ponsford, Jennie; Hicks, Amelia; Leon-Novelo, Luis; Ngan, Esther; Sander, Angelle M.
• Is Part Of: Journal of head trauma rehabilitation. Volume 32:Issue 5(2017)
• Publication Date: 2017-09
• Subjects: classification -- cross-validation -- traumatic brain injury; Brain damage -- Patients -- Rehabilitation -- Periodicals; Brain damage -- Periodicals; 617.4810443
• DOI: 10.1097/HTR.0000000000000290
• Languages: English
• ISSNs: 0885-9701

Abstract

Abstract : Objective: To (1) identify groupings of persons with traumatic brain injury (TBI) in the posthospital period in a cohort recruited in Australia; (2) compare groupings from the Australian cohort to groupings previously developed for a US cohort. Setting: Rehabilitation centers in the US and Australia. Participants: A total of 170 persons with TBI from Australia and 504 persons with TBI from the United States. Participants were aged 18 to 64 years, greater than 6 months postinjury, had capacity to give consent, and had definitive contemporaneous medical evidence of TBI. Design: Observational study. Main Measures: Cognitive tests—Wechsler Letter-Number Sequencing and Coding, Rey Auditory Verbal Learning Test, Trail Making Test, Verbal Fluency. Questionnaires—9 scales from the Traumatic Brain Injury Quality-of-Life system; Neurobehavioral Symptom Inventory; Economic Quality of Life, Family Assessment Device—General Functioning. Performance validity—Word Memory Test. Results: Agreement in classification for the 2 samples was only moderate with 63.5% correctly classified and Cohen's κ = 0.53. A post hoc analysis placing all persons showing poor performance validity in the same group improved classification to 73% and Cohen's κ = 0.65 indicating substantial agreement. Conclusion: Results provided support for the patient groups developed for the US sample and indicate that these groupings largely replicated in a new cohort.